History
2025: Hygon GPUs beat Nvidia chips by 10x
In mid-February 2025, specialists from Hokhai University in Nanjing (PRC) announced that they had managed to create a system based on Chinese GPUs (GPUs), which in performance is about 10 times higher than similar platforms with advanced Nvidia accelerators .
Powerful Nvidia GPU-based cards are widely used to solve resource-intensive tasks in the field of artificial intelligence, as well as to create supercomputer complexes. However, China's access to such products is limited due to US sanctions. Therefore, local scientists and engineers are actively designing their own products that can become an alternative to Western products.
Professor Nan Tongchao of Hohai University, along with his team, focused on using Chinese CPUs CPU and GPUs to develop a more efficient supercomputer architecture. Experts came to the conclusion that the key to minimizing performance losses in parallel computing is efficient data transfer and coordination of tasks between system nodes. To improve the performance of the GPU platform, the project participants created special software optimization tools.
The new system uses Chinese x86-compatible Hygon 7185 chips (32 cores, 64 streams, 2.5 GHz) and graphics accelerators developed in China as hardware components. Network infrastructure bandwidth - 200 Gb/s. Experiments have shown that the proposed approach allows you to increase performance by about an order of magnitude when solving certain problems, for example, when predicting floods. The achievement is said to challenge the dominance of American-made graphics chips in scientific computing.[1]
Notes
| The site content is translated by machine translation software powered by PROMT. The machine-translated articles are not always perfect and may contain errors in vocabulary, syntax or grammar. Read original article If you find inaccuracies or errors in the results of machine translation, please write to editor@tadviser.ru. We will make every effort to correct them as soon as possible. |